Agentic WeeklyGet the weekly

LLM API pricing comparison, October 2026: USD per 1M tokens

As of our reads on Oct 26, 2026, 41 language models in our data have a USD list price. The cheapest by input plus output is Ministral 3 14B at $0.20 / $0.20 per million tokens; the most expensive is GPT-5.5 Pro at $30 / $180. For coding, the cheapest models in the top 10 of the LMArena WebDev board (Oct 22, 2026) are Claude Sonnet 5.5, GPT-6.1 Sol, Kimi K3.1 Pro and GPT-6 Sol at $2 / $10 per million tokens. Every row below links to the vendor page; check it before you budget.

By Agentic Weekly, operated independently from Texas, US · Last updated (America/Chicago) · How we work

Free weekly brief: email newsletter · RSS

Evidence: Reported. Prices are USD list prices per 1M tokens, standard processing, read Oct 26, 2026; batch, flex and fast-mode multipliers are not in the table read from the sources listed below on 2026-10-27 (America/Chicago). We did not run these products or models for this page.

Cite this: Agentic Weekly, "LLM API pricing comparison, October 2026: USD per 1M tokens", https://theagenticweekly.com/what-is/llm-pricing-comparison, read Oct 27, 2026.

Method: USD list prices per 1M tokens, standard processing; batch, flex and fast-mode rates are excluded. Each fact is copied from the vendor page linked in its row into our model records with the date we read it. Reported means read from that page, not run by us; Tested would mean we ran it and logged a dated run, and no model on this page is Tested. 64 language models are in our data: 41 have a sourced USD list price, 2 are priced in another currency and 21 have no verified price. Only models with a sourced price are listed; we do not estimate or pad.

Data: models.csv · models.json · answers.json

Price table: 41 models, USD per 1M tokens

Sorted by output price, lowest first. "Cached input" is the vendor's discounted rate for repeated prompt prefixes. The Read column is the date we read that vendor page.

LLM API list prices in USD per 1M tokens, read 2026-10-26
ModelVendorInputOutputCached inputNotesSourceRead
Ministral 3 14BMistral AI$0.20$0.20n/anonevendor page2026-10-26
Mercury 3Inception$0.10$0.35n/anonevendor page2026-10-26
Claude Haiku 5.5Anthropic$0.10$0.50n/anonevendor page2026-10-26
GPT-6 LunaOpenAI$0.10$0.50$0.01nonevendor page2026-10-26
MiniMax M3.1 FlashMiniMax$0.15$0.60$0.015nonevendor page2026-10-26
Mistral Small 4Mistral AI$0.15$0.60n/anonevendor page2026-10-26
GPT-5.6 LunaOpenAI$0.20$1.20$0.02nonevendor page2026-10-26
Grok Build 0.1xAI$1$2$0.20limited accessvendor page2026-10-26
Grok 4.20xAI$1.25$2.50$0.20nonevendor page2026-10-26
Gemini 3.8 FlashGoogle$0.75$3.75n/afrom 2027-01-01: $1.50 / $7.50vendor page2026-10-26
Mistral Large 4Mistral AI$1.36$4.18$0.14public previewvendor page2026-10-26
GLM-5.3Z.ai$1.40$4.4$0.26nonevendor page2026-10-26
GPT-5.4 miniOpenAI$0.75$4.50$0.075nonevendor page2026-10-26
GPT-6.1 TerraOpenAI$1$5$0.05Prompts above 272K input tokens: $2 / $7.50vendor page2026-10-26
Grok 4.5xAI$2$6$0.30nonevendor page2026-10-26
Grok 4.6xAI$2$6$0.50nonevendor page2026-10-26
Grok 4.7xAI$2$6$0.50Prompts above 200K tokens: $4 / $12vendor page2026-10-26
Mistral Medium 3.5Mistral AI$1.50$7.50n/anonevendor page2026-10-26
Claude Sonnet 5Anthropic$2$10n/anonevendor page2026-10-26
Claude Sonnet 5.5Anthropic$2$10n/anonevendor page2026-10-26
Gemini 4 ArgonGoogle$2$10$0.10from 2026-12-01: $4 / $20vendor page2026-10-26
GPT-6 SolOpenAI$2$10$0.20nonevendor page2026-10-26
GPT-6.1 SolOpenAI$2$10$0.10nonevendor page2026-10-26
Kimi K3.1 ProMoonshot AI$2$10$0.20nonevendor page2026-10-26
GPT-5.6 TerraOpenAI$2$12$0.20nonevendor page2026-10-26
GPT-5.4OpenAI$2.50$15$0.25nonevendor page2026-10-26
Kimi K3Moonshot AI$3$15$0.30nonevendor page2026-10-26
Claude Opus 5.5Anthropic$4$20n/anonevendor page2026-10-26
GPT-5.6 SolOpenAI$4$20$0.40nonevendor page2026-10-26
Kimi K3.1 Pro HighspeedMoonshot AI$4$20$0.20nonevendor page2026-10-26
Muse WonderMeta$4$20$0.20public previewvendor page2026-10-26
Claude Opus 4.6Anthropic$5$25n/anonevendor page2026-10-26
Claude Opus 4.7Anthropic$5$25n/anonevendor page2026-10-26
Claude Opus 4.8Anthropic$5$25n/anonevendor page2026-10-26
Claude Opus 5Anthropic$5$25n/anonevendor page2026-10-26
GPT-5.5OpenAI$5$30$0.50nonevendor page2026-10-26
Claude Fable 5Anthropic$10$50n/anonevendor page2026-10-26
Claude Fable 5.1Anthropic$10$50n/anonevendor page2026-10-26
Claude Mythos 5.1Anthropic$10$50n/alimited accessvendor page2026-10-26
GPT-6 AstraOpenAI$10$50n/anonevendor page2026-10-26
GPT-5.5 ProOpenAI$30$180n/anonevendor page2026-10-26

Prices not in US dollars

These vendors price in Chinese yuan. We do not convert currencies, so they are left out of the USD ranking above.

LLM API list prices in CNY per 1M tokens
ModelVendorInputOutputCached inputNotesSourceRead
DeepSeek-V4.1-Flash-1017DeepSeekCNY 1CNY 4CNY 0.02peak hours: CNY 2 / 8vendor page2026-10-26
DeepSeek-V4.1-FlashDeepSeekCNY 1CNY 4CNY 0.02peak hours: CNY 2 / 8vendor page2026-10-26

Models with no price recorded

21 language models in our data have no verified price yet, so they are not ranked: Claude Mythos 5, Claude Sonnet 4.6, Command A+, Gemini 3.1 Flash-Lite, Gemini 3.1 Pro, Gemini 3.5 Flash-Lite, Gemini 3.6 Flash, Gemini Robotics ER 2, Gemma 4, GLM-5.1, GLM-5.2, GLM-5.3-Flash, GLM-5, Muse Spark Open, Muse Wonder Full, Nemotron 3.5 Bolt E4B (4GB), North Mini Code, North Small Translate, Qwen3.6-27B, Qwen3.8-2.4T-A95B, Qwen3.8-27B.

How to read this table

  • List price is not total cost: output-heavy tasks cost more than the input price suggests, and reasoning models can spend many output tokens.
  • Tier and peak-hour notes come from the same vendor page as the row; a later price (for example a scheduled change) is shown with its start date.
  • Cheaper per token is not better per task. We have not measured cost per task for these models.
  • For the cheapest options and the cheapest for coding, see cheapest LLM API.

FAQ

How much does the Claude API cost per million tokens?

Claude Fable 5.1 $10 / $50; Claude Opus 5.5 $4 / $20; Claude Sonnet 5.5 $2 / $10; Claude Haiku 5.5 from $0.10 / $0.50 (USD per 1M input / output tokens, read Oct 26, 2026, Anthropic pricing). Cheaper cached and batch rates exist; see each model page.

Which LLM API is the most expensive?

GPT-5.5 Pro at $30 / $180 per 1M tokens, the highest output price among the 41 priced models in our data (read Oct 26, 2026).

Are DeepSeek prices in this table?

Not in the USD ranking. DeepSeek prices in Chinese yuan (DeepSeek-V4.1-Flash-1017: CNY 1 input / CNY 4 output per 1M tokens), and we do not convert currencies.

How often are these prices updated?

Each row shows the date we read the vendor page (latest Oct 26, 2026). Facts older than 14 days are flagged for re-checking on the models page.

Sources

Page history

  • : First published. Facts read from the sources listed below on this date.

Markdown version for agents: /what-is/llm-pricing-comparison.md